Blade servers are widely used in datacenters to save space and improve system management. They are self-contained computer servers, designed for high density. Blade servers have many components removed for space, power and other considerations while still having all the functional components to be considered a computer.
One of the principal benefits of blade computing is that components are no longer restricted to the minimum size requirements of the standard server-rack configuration. For example, densities of 100 computers per rack and more are achievable with the present blade systems. Furthermore, as more processing power, memory and I/O bandwidth are added to blade servers, they are being used for larger and more diverse workloads.
However, there exist many limitations to the current state of technology with respect to blade servers. For example, blade server configurations are fixed and only communicate through switch I/O fabrics. This method of communication does not allow for high performance while hosting a large scale symmetric multiprocessing system. Furthermore, while high end systems have the ability to aggregate cell boards by using proprietary switch I/O fabrics driven by custom ASICs and fixed large scale mid-planes, this procedure is cost prohibitive.